Lithiumniobate Die Assembled by a Low-stress Soldering Technique
Method to Fasten a Surface Acoustic Wave Sensor
Pol Ribes-Pleguezuelo
1,2
, Katherine Frei
1
, Gudrun Bruckner
3
, Erik Beckert
1
,
Ramona Eberhardt
1
and Andreas Tünnermann
1,2
1
Fraunhofer Institute for Applied Optics and Precision Engineering IOF, Albert-Einstein-Str. 7, 07745 Jena, Germany
2
Institute of Applied Physics IAP, Friedrich-Schiller University Jena, Max-Wien-Platz 1, 07743 Jena, Germany
3
Carinthian Tech Research AG, Europastraße 12, A-9524, St. Magdalen, Austria
Keywords: Surface Acoustic Wave Sensor, Optical Crystal, Packaging, Solderjet Bumping Technique, Lithiumniobate.
Abstract: Solderjet bumping technique was applied to assemble a Surface Acoustic Wave (SAW) sensor prototype
designed with a lithiumniobate crystal and a base sub-mount made of stainless steel. The assembly was
designed with this technology in order to withstand the device’s mechanical strength requirements. The initial
performed tests showed that the solderjet bumping technique can be used to assemble brittle components
without creating internal damage in the crystal. The selected solder alloy Au
80
Sn
20
used to fasten the
lithiumniobate showed proper alloy wettability and joint strength on the crystal and on the substrate material.
Finally, a lithiumniobate die device was soldered by soldering means to the stainless steel sub-mount, and
withstood the strength device requirements by passing robustness (push) tests.
1 INTRODUCTION
1.1 Motivation
The demand from industry and consumer electronics
for continuous miniaturisation and increased
integration density of components can only be
fulfilled by applying appropriate attachment and
interconnection technologies. Smart systems require
interconnection of heterogeneous materials like
crystals, ceramics, printed circuit boards, screen
printed layers, and metals. The die attachment
processes are critical for the functionality of the
devices. Depending on the application, the inter-
connection layer must act as a thermal or electrical
contact, allowing power dissipation or compensating
thermal mismatch of different materials.
Lithiumniobate (LN) is a material widely used in
optics and as High Frequency (HF) filters. This piezo
electric crystal is also used for Surface Acoustic
Wave (SAW) based physical sensors, which work
completely passively without power supply, and
can be interrogated wirelessly and used in temperatu-
re regions well above the operation range of silicon
based devices (Reindl, 1998; Bruckner, 2017). The
high temperature operation (300 °C) poses additional
demands on the interconnection techniques.
In this work, we examine the solderjet bumping
technique for the die attachment of LN chips to
stainless steel, which promises several advantages
when compared to common adhesives based on
organic materials, as these materials are generally
limited to temperatures below 280 °C. The final goal is
to establish a bonding process that permits LN crystals
to function as combined temperature and strain or
pressure sensors in a wide temperature range (-100 °C
to 300 °C) with high reproducibility and reliability.
SAW chips are commonly fixed using silicon
based rather soft adhesives which allow low stress
bonds, but are limited to ~ 200°C. During the last
years we have tested high temperature stable epoxy-
and polyimide based glues for the attachment of SAW
dies in metallized and ceramic packages. All of these
adhesives failed in long term tests where the samples
were exposed to temperatures around 300°C for
several months. We observed outgassing that resulted
in contaminating layers on the SAW chips and
disintegration on the adhesives. In extreme cases the
glues disintegrated completely and the dies fell off.
Any outgassing from the adhesives or package can
raise chemical reaction in the microclimate of
packaged devices which damage the delicate surface
of SAW chips (Bardong 2016). Ceramic glues could
be used, but the TCE mismatch often causes breaking
Ribes-Pleguezuelo, P., Frei, K., Bruckner, G., Beckert, E., Eberhardt, R. and Tünnermann, A.
Lithiumniobate Die Assembled by a Low-stress Soldering Technique - Method to Fasten a Surface Acoustic Wave Sensor.
DOI: 10.5220/0006645400910097
In Proceedings of the 6th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2018), pages 91-97
ISBN: 978-989-758-286-8
Copyright © 2018 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
91
of the chips. Metal based attachment techniques are
promising for high temperature and low stress die
attachment (Roshanghias 2017) as they allow
compensation of TCE mismatch and minimize
mechanical stress. Solderjet bumping technique is
especially suited, as no thermal stress is induced
during the bonding process.
1.2 Solderjet Bumping Technique
Solder-joining using metallic solder alloys is an
alternative to adhesive bonding. Laser-based
soldering processes are especially well suited for the
joining of optical components made of fragile and
brittle materials such as glasses, ceramics and
crystals, due to a localized and minimized input of
thermal energy (Beckert, 2009).
Solderjet bumping (Figure 1) is a technique
adapted from flip chip processing of semiconductor
devices that also allows for the flux-free and contact-
free processing of optical components and 3D-
packaging. It uses spherical solder preforms of
various soft solder alloys (e.g. tin-based lead-free
solders, low melting indium alloys or high melting
eutectic gold-tin, gold-silicon or gold-germanium
solders) in a diameter range of 40 µm to 760 µm. The
solder spheres are transferred from a reservoir to a
placement capillary with a conical tip and an inner
diameter that is slightly smaller than the diameter of
the spheres. After positioning the capillary next to the
joining geometry using an articulating robot, the
solder alloy is molten by an infrared laser pulse and
jetted out of the capillary by applying nitrogen
pressure. The jetting of liquid solder volumes
provides a very good thermal contact of the alloy with
the components, and allows for the joining within
complex 3D-integrated geometries. The bond head of
the solderjet bumper integrates solder volume
feeding, reflow, and application of liquid solder and
allowing for highly automated and flexible use.
Figure 1: Schematic drawing of the solderjet bond head able
to solder droplets with 6 degrees of freedom (DOF).
However, the formation of a metallic solder joint
using components made of non-metallic materials
with solderjet bumping requires a wettable
metallization layer applied to the components. Such
surfaces can be provided by thin film (e.g. physical
vapour deposition) or thick film (e.g. screen printing
of metal pastes) processes. Sputtered three-layer
systems (Figure 2) using a titanium adhesion layer, a
platinum diffusion barrier, and a noble gold finish
preventing oxidization and acting as a wetting
surface, provide superb conditions for wetting of
liquid solder droplets (Banse, 2005).
Figure 2: Example of three sputtered layers (Ti/Pt/Au) over
the substrate component to later be able to create a wettable
surface for the solder bump.
2 REQUIREMENTS
As stated before, the aimed operation temperature is
300 °C or higher while the processing temperature for
LN is limited to about 450 °C for a short time. Die
attachment by solderjet bumping with eutectic
materials allows high temperature operation of the
devices with minimum thermal stress induced during
the attachment process. In addition, metals are ductile
and can compensate thermal mismatch of different
materials. This property is extremely important for
LN, as the crystal is highly anisotropic. The thermal
expansion coefficient in one direction is about three
times bigger than in the other, preventing thermal
matching to metal or ceramics. Furthermore, the
metallic interface allows a good thermal contact
between chip and support, which is crucial for the
device to operate as temperature sensor.
For sensing of pressure or strain several
geometries can be used. We have considered a
cantilever setup, where the LN crystal is either
mounted on a metallic cantilever beam, or forms the
cantilever itself. In both configurations precise and
reproducible mounting of the device with sufficient
adhesion strength is essential. When solderjet
bumping is used, the die can be put in direct contact
with the metal of the pressure sensor. This is a big
advantage compared to any attachment technique
PHOTOPTICS 2018 - 6th International Conference on Photonics, Optics and Laser Technology
92
using an interface layer, as any adhesive within the
chip- metal interface would affect the device
sensitivity.
Figure 3 shows a sketch of such a sensor. The size
of the chip used for the experiments was 8 mm x 3 mm
x 0.35 mm. LN was chosen as substrate for the sensor
element because of the high coupling coefficient that
provides a high freedom in sensor design and linear
sensitivity to strain (Bruckner, 2013).
Figure 3: Schematics for a pressure sensor with thin
membrane and chip as cantilever.
3 ASSEMBLY EXPERIMENTAL
DETAILS
For this study, the solderjet bumping laser energy
parameters (represented by varying the laser
current (mA) and the laser pulse (ms)) were tuned to
provide the optimal wetting and strength of Au
80
Sn
20
100 µm diameter spherical solder-bumps on a
metalized LN substrate. Au
80
Sn
20
was selected due to
its melting temperature, low enough not to damage
the crystal during soldering alloy reflow, but high
enough to resist temperatures close to 300 °C required
for the final device application. The goal was to
successfully assemble a LN crystal to a stainless steel
baseplate. The study was done by following a design
of experiments (DoE) using input factors, the laser
current (mA) and laser pulse (ms), to study the
material responses: i) damage to the material, ii)
correct wetting provided by a correct bump melting
diameter, and iii) force needed to shear the bumps.
An important factor in this assessment was to
avoid damage to the LN substrate. A common
problem with use of LN is its high tendency to receive
photorefractive damage, which results in a change in
inhomogeneity of the refractive index of the material
(Levinstein, 1967). Another important factor is the
difference of strength and wetting of Au
80
Sn
20
solder
on both LN and stainless steel by applying bumps
using the same current and pulse energy, which is
attributed to their respective thermo-mechanical
characteristics. The relevant material characteristics
of the substrate material, the LN and the selected
alloy are listed in Table 1.
Table 1: Thermo-mechanical properties of used materials.
Lithium-
Niobate
Stainless
Steel
Au
80
Sn
20
Density (g/cm3)
4.64
8
14.7
Young's Modulus
(GPa)
170
195
68
Thermal Conductivity
(W/mK)
4.4
16.2
57,0
Melting Temp (°C)
--
--
280
3.1 Initial Tests
Preliminary tests were performed to determine the
maximum and minimum laser energy parameters.
The maximum energy was determined when the
energy was the highest possible without damaging the
substrate, while the minimum was determined when
the energy was lowest while still allowing flow of the
solder bump through the solderjet bumping jet
capillary (Figure 1). After preliminary tests were
done to find the maximum and minimum energy
parameters, a DOE was designed by using twenty-six
different laser energy points represented by their laser
current (mA) and laser pulse (ms).
The initial created damage was analyzed through
visual inspection, rating the damage on a scale of 1 to
4. (Figure 4).
Figure 4: Example of examined damage. (1) No presence of
damage; (2) material abrasion; (3) crack; (4) major crack.
Lithiumniobate Die Assembled by a Low-stress Soldering Technique - Method to Fasten a Surface Acoustic Wave Sensor
93
The diameter and shear strength results are
displayed graphically with a contour map on Figure 5.
In Figure 5, the damage on the LN substrate is shown
in red numbers at each point on the map (Upper
image).
Figure 5: Results of DoE. Upper image, bump diameter and
in red numbers damage to assess correct wettability. Image
below, bump shear strength results.
The ideal diameter of solder bump after reflow
should be approximately 175% to 200% of the
original solder sphere (Mäusezahl, 2016); for the case
of 100 µm it should be about 175 to 200 µm. As can
be seen in Figure 5, there are some results that have
an approximate diameter size of 200 µm, in particular
bumps with laser energy parameters of
3000 mA/0.5 ms, 3866 mA/0.5 ms, and
4733 mA/0.4 ms. These samples also showed no
damage and good resulting shear strength.
The tensile strength of Au
80
Sn
20
is 275 MPa
(Indium Corp., 2013) and typically shear strength
within metals is estimated to be 50% to 60% of the
ultimate tensile strength (Callister, 2013). For our
case, this results in an estimation of 137.8 to
165 MPa. In Figure 5, it can be seen that in cases with
very high laser energy, shear strength was much
below this estimation, however at lower energies
shear strength was still slightly above this estimation.
Figure 6 shows an example of a sheared solder bumps
using low energy (left) and high energy (right). A
possible reason for the weakening of Au
80
Sn
20
, is the
formation of AuSn
4
, a common occurrence when
soldering with tin and gold materiel. This compound
is known to weaken the material and the joint. Higher
energy may allow easier formation of AuSn
4
(Hare,
2010).
Figure 6: Shear test examples performed using 100 µm
AuSn bumps. Left, low energy similar to 8.9 mJ by using a
solderjet bumping settings of 3000 mA and 0.5 ms. Right,
high energy similar to 15 mJ by using a solderjet bumping
settings of 3000 mA and 0.8 ms.
These energies were also tested on stainless steel,
the base metal in which the LN is to be bonded to.
Examples of the results are shown in Figure 7. The
shear strength and wetting diameter values for these
samples are shown to be much higher than those that
were soldered on the layered LN substrate. The
reason for this may be due to the significantly
different thermal conductivity in LN and stainless
steel. Stainless steel’s higher thermal conductivity
allows for a larger spread of heat within the metal
during the period that the laser contacts the surface.
This larger heated area allows for easier wetting of
Au
80
Sn
20
solder. In LN, its low conductivity means
that heat will remain concentrated in a small area,
meaning that this area may achieve a higher
temperature than that of stainless steel, during laser
contact. As we have seen from the design of
experiments, higher laser energy, which correlates to
higher temperature, results in lower shear strength.
This may explain the reason why LN solder samples
show lower wetting diameters and lower shear
strength. The optimized laser energy was finally
obtained by using the laser values of 3000 mA and
0.5 ms (representing and approximate energy of
8.9 mJ).
PHOTOPTICS 2018 - 6th International Conference on Photonics, Optics and Laser Technology
94
Figure 7: Similar selected energy tests on stainless steel. (1)
3000 mA/0.5 ms showed a bumped diameter of 293.5 µm
and a shear force of 406 MPa. (2) 3866 mA/0.5 ms showed
a bumped diameter of 171.7 µm and a shear force of
235 MPa. (3) 4733 mA/0.4 ms showed a bumped diameter
of 172.5 µm and a shear force of 301 MPa.
The final selected laser parameters (3000 mA and
0.5 ms) were used to create a prototype LN and
stainless steel assembly which was later tested for its
strength and durability.
3.2 Assembly Procedure
Pac Tech’s Solder Ball Bumper SB2-Jet was used to
solder 100 µm AuSn bumps onto the surface of
metalized LN substrate and stainless steel base.
Using the selected optimal laser energy of 8.9 mJ
represented by using and 3000 mA and 0.5 ms, LN
samples were soldered to stainless steel with the
following methods. The LN samples were
temporarily secured to stainless steel to later be
rotated by 45°, resulting in the desired jointing edges
facing up toward the solderjet bumping capillary
(Figure 8(a)). Two methods were used to create a
joint between the LN and stainless steel. In the first
method, a rectangular pattern of four by ten bumps
was placed in four locations along each long edge of
the LN sample. These were placed in a way that half
the solder bump pattern bonded with the LN and the
other half bonded with the stainless steel (Figure
8(b)). The second method was similar to the first,
except that the bump pattern was only placed in three
locations per side. This pattern was also overlaid,
with one on top of the other (Figure 8 (c)) to secure
the bond between bumps.
Figure 8: Schematic drawing of different soldering
approaches of 100 µm solder bumps application. (a)
Schematic of the process with the laser reflow being applied
onto the bumps. (b) First described method. (c) Second
described method.
Following this method, several prototypes were
assembled as described (Figure 9).
Figure 9: Finally assembled samples.
4 ENVIRONMENTAL TESTING
4.1 Push Tests
With the selected laser energy, the LN was secured
onto stainless steel using the two methods described
in the Assembly Procedure section (Figure 8 and
Figure 9). This was then subject to a push test. The
graphical results of the push test for methods 1 and 2
are shown in Figure 10. The first method showed a
maximum strength of 18.1 N. The graph of the second
method shows a spike occurring at 18.8 N, and then
suddenly dropping to zero, then rising again and
remaining steady at about 3 N. The initial spike
represents the force in which the LN broke from the
soldered joint, and the drop represents a sudden slip
after failure. The steady 3 N force indicates the LN
continued to slide between the solder joints.
Comparing the results of these two methods, it seems
that the second method is slightly better than the first
with a higher failure force. The disadvantage of this
method however is that more time and solder alloy is
required. Considering this and the small difference of
0.7 N force observed between the two methods, the
Lithiumniobate Die Assembled by a Low-stress Soldering Technique - Method to Fasten a Surface Acoustic Wave Sensor
95
first assembly method was selected as the most
optimal.
Figure 10: The graphical results of the push test for
methods 1 and 2. The graphic represents the moving
distance of the Push tests machine (Zwick Roell Z020) and
the derivate force applied onto the LN assembled crystal.
4.2 Thermal Tests
Thermal annealing tests were performed on four
samples at 300 °C to prove the devices operation
requirements. First the temperature was slowly
increased from room temperature at a rate of
100 °C/h, to finally be kept at a constant 300 °C for
12 h. Afterwards the samples were optically
inspected to check for cracks of the LN chips induced
by thermal stress. As all samples passed the test
successfully they were exposed to a strong
temperature gradient by putting the devices from
room temperature directly into a hot environment at
300 °C and out again. All devices could withstand
three cycles without cracks. Further test are planned
with larger number of samples applying different
temperatures.
5 CONCLUSIONS
Both, the optimal shear strength and wetting
parameters to assemble the lithiumniobate crystal to
a stainless steel sub-mount were shown to be at
3000 mA and 0.5 ms. The general pattern is that
lower laser energies results in higher shear strength
and smaller wetting diameters for the solder bumps.
Soldering Au
80
Sn
20
to stainless steel results in larger
wetting diameters and higher shear strength than
when soldered to lithiumniobate, likely due to the
higher thermal conductivity of stainless steel.
The two assembly methods produced remarkably
similar results in their strength, however considering
that the second method uses more material and time
to be performed, the optimal method is chosen to be
the first one. The final selected method showed to be
capable of handling the required strength demands
imposed by the final device application.
Moreover, other author’s publications focused on
the study of the birefringence effects produced by the
described soldering procedure showed how solderjet
bumping process can assemble optical components
with just a residual stress without affecting the device
optical performances (Ribes-Pleguezuelo, 2016;
Ribes-Pleguezuelo, 2017). Previous studies and
publications make us consider that the same residual
effect will not alter the device functionality for the
present assembled devices.
ACKNOWLEDGEMENTS
The authors want to acknowledge other members of
the Fraunhofer IOF for their support, especially to
Marcel Hornaff for the provided help with the DOE.
To the Deutscher Akademischer Austauschdienst
(DAAD).
This project is partly supported within the
COMET Competence Centers for Excellent
Technologies - Program by BMVIT, BMWFJ and the
Province of Carinthia.
REFERENCES
Banse, H., Beckert, E., Eberhardt, R., Stöckl, W., Vogel, J.,
2005. Laser Beam Soldering a New Assembly
Technology for Microoptical Systems. In Microsystems
Technologies 11, pp. 186-193.
Bardong, J., Binder, A., Toskov, S., Miskovic, G.,and
Radosavljevic, G., 2016. Investigation of low-tempera-
ture cofired ceramics packages for high-temperature
SAW sensors. In J. Sens. Sens. Syst., 5, p 8593.
Beckert, E., Oppert, T., Azdasht, G., Zakel, E., Burkhardt,
T., Hornaff, M., Kamm, A., Scheidig, I., Eberhardt, R.,
Tünnermann, A., Buchmann, F., 2009. Solder Jetting -
A Versatile Packaging and Assembly Technology for
Hybrid Photonics and Optoelectronical Systems. In
IMAPS 42nd International Symposium on
Microelectronics, Proceedings, Vo. 42, pp. 406-412.
Bruckner, G., Schicker, J., Schlumpf, P., 2013, Stress
sensitivity of SAW Rayleigh waves on Lithiumniobate
and its application in pressure sensor design.
Proceedings IEEE International Ultrasonics
Symposium (IUS), Prague, pp. 2140-2143.
Bruckner,G., Bardong, J., Binder, A., Nicolay, P., 2017,
SAW Delay Lines as Wireless Sensors for Industrial
Applications. Proceedings of the VIII ECCOMAS
Thematic Conference on Smart Structures and
Materials, SMART 2017, Madrid, pp. 1433-1442.
PHOTOPTICS 2018 - 6th International Conference on Photonics, Optics and Laser Technology
96
Callister, W. D., Rethwisch D. G., 2013. Materials Science
and Engineering: an Introduction. Wiley, New Jersey,
9
th
edition.
Hare, E., 2010. Gold Embrittlement of Solder Joints. In
SEM Lab, INC. www.semlab.com
Indium Corporation, 2013. Gold-Tin: The Unique Eutectic
Solder Alloy. In www.documents.indium.com/
qdynamo/download.php?docid=2103.
Levinstein, H. J, Ballman, A. A., Denton, R. T., Ashkin, A.,
Dziedzic, J.M., 1967. Optically Induced refractive
index inhomogeneities in LiNBo3 and LiTaO3. In
Applied Physics Letters, vol. 9, no. 1, 1966, pp. 7274;
doi:10.1063/1.1754607.
Mäusezahl, M., Hornaff, M., Burkhardt, T., Beckert, E.,
2016. Mechanical Properties of Laser-Jetted SAC305
Solder on Coated Optical Surfaces. In Physics
Procedia, vol. 83, 2016, pp. 532539;, doi:10.1016/
j.phpro.2016.08.055.
Reindl, L., Scholl, G., Ostertag, T., Scherr, H., Wolff, U.,
Schmidt, F., 1998, Theory and Application of Passive
SAW Radio Transponders as Sensors. IEEE Trans.
Ultrason. Ferroelectr. Freq. Control 45 (5), pp.1281-
1292.
Ribes-Pleguezuelo, P., Koechlin, C., Hornaff, M., Kamm,
A., Beckert, E., Fiault, G., Eberhardt, R., Tünnermann,
A., 2016. High-precision optomechanical lens system
for space applications assembled by a local soldering
technique. In Opt. Eng. 55(6), 065101 (2016); doi:
10.1117/1.OE.55.6.065101.
Ribes-Pleguezuelo, P., Zhang, Z., Beckert, E., Eberhardt,
R., Wyrowski, F.,Tünnermann, A., 2017. Method to
simulate and analyse induced stresses for laser crystal
packaging technologies. In Optics Express Vol. 25,
No.6; doi:10.1364/OE.25.005927
Roshanghias, A., Bruckner, G., Binder, A.,2017. High
temperature MEMS packages: die-attach solutions for
LiNbO3 under low bonding pressures. In J Mater Sci:
Mater Electron, DOI 10.1007/s10854-017-6605-3
Lithiumniobate Die Assembled by a Low-stress Soldering Technique - Method to Fasten a Surface Acoustic Wave Sensor
97